Spatial models of terrestrial biogeochemistry: Field and model studies of Douglas-fir forests

Due to the rapid rate at which humans are altering their environment, it is necessary to understand how environmental change alters biogeochemical processes, while our understanding of the basic processes is still incomplete. Modeling studies can guide the collection of process data by outlining which processes and properties control system-wide model dynamics. However, modeling studies are constrained by the absence of spatially explicit datasets of terrestrial biogeochemistry. As a result, biogeochemical models have focused on describing mean dynamics, generally describing ecosystems using averaged biogeochemical parameters, and thus ignoring biogeochemical variation.; This dissertation is motivated by the widespread application of aggregated biogeochemical models. Given the increasing use of models as predictive and policy-making tools, it is crucial to understand the effect of applying aggregated biogeochemical models to ecosystems known to have measurable variation in biogeochemical properties. My dissertation tests the value of accounting for spatial heterogeneity in biogeochemical models of forest systems by combining the analysis of spatial field data with the development of models.; The distribution of carbon and nitrogen in leaf, litter, and soil was documented for managed and old-growth Douglas-fir forests. While there was widespread variation in forest biogeochemical properties, a specific spatial pattern did not emerge in the biogeochemical data.; Three models of nitrogen cycling were developed and analyzed for their sensitivity to parameter variation and their response to perturbation. All models produced realistic values for plant, soil, and microbial nitrogen pools, though they differed in their response to perturbations simulating forest management. Parameter sensitivity analysis showed that the plant uptake half-saturation constant dominated the dynamics of all model systems studied.; Spatial models were developed using diffusion to approximate nitrogen transport. Carbon and nitrogen distributions measured in the field research were used to study the effect of aggregating heterogeneity on model dynamics. The maintenance of nutrient heterogeneity was controlled by nutrient transport assumptions. When soil N transport dominated transport dynamics, nitrogen heterogeneity was significant. However if strong plant and soil nitrogen transport coincided, stand-scale N dynamics rapidly homogenized. While the spatial resolution minimally affected stand nitrogen means, the scale of spatial resolution significantly affected nitrogen variance. For all models, the reaction terms dominate mean stand N dynamics, while diffusion terms control nitrogen heterogeneity.; Under model application to forest management, 40-year rotations led to long-term declines in forest productivity, regardless of post-harvest management. In contrast, 80-year rotations sustained long-term productivity providing that post-harvest management included retention of organic matter.